CONF Madikeri_ICASSP2019_2019/IDIAP A BAYESIAN APPROACH TO INTER-TASK FUSION FOR SPEAKER RECOGNITION Madikeri, Srikanth Dey, Subhadeep Motlicek, Petr bayesian fusion inter-task fusion speaker recognition EXTERNAL https://publications.idiap.ch/attachments/papers/2019/Madikeri_ICASSP2019_2019.pdf PUBLIC https://publications.idiap.ch/index.php/publications/showcite/Madikeri_Idiap-RR-07-2020 Related documents In Proceedings of ICASSP 2019 Brighton, ENGLAND 2019 5786-5790 1520-6149 978-1-4799-8131-1 In i-vector based speaker recognition systems, back-end classifiers are trained to factor out nuisance information and retain only the speaker identity. As a result, variabilities arising due to gender, language and accent ( among many others) are suppressed. Inter-task fusion, in which such metadata information obtained from automatic systems is used, has been shown to improve speaker recognition performance. In this paper, we explore a Bayesian approach towards inter-task fusion. Speaker similarity score for a test recording is obtained by marginalizing the posterior probability of a speaker. Gender and language probabilities for the test audio are combined with speaker posteriors to obtain a final speaker score. The proposed approach is demonstrated for speaker verification and speaker identification tasks on the NIST SRE 2008 dataset. Relative improvements of up to 10% and 8% are obtained when fusing gender and language information, respectively. REPORT Madikeri_Idiap-RR-07-2020/IDIAP A BAYESIAN APPROACH TO INTER-TASK FUSION FOR SPEAKER RECOGNITION Madikeri, Srikanth Dey, Subhadeep Motlicek, Petr EXTERNAL https://publications.idiap.ch/attachments/reports/2018/Madikeri_Idiap-RR-07-2020.pdf PUBLIC Idiap-RR-07-2020 2020 Idiap March 2020